On the invertibility of matrices with a double saddle-point structure (Q6615432)

Let us consider the following \(\ell \times \ell\) double saddle-point matrix in the form of a symmetric three-by-three block matrix \[\mathcal{K}=\begin{bmatrix} A & B^T &0 \\\NB & -D & C^T\\\N0 & C& E \end{bmatrix},\] where \(A \in \mathbb{R}^{n\times n}\), \(D \in \mathbb{R}^{m\times m}\), \(E \in \mathbb{R}^{p\times p}\), and \(\ell = n + m + p\). These matrices arise in various applications in computational science and engineering. \textit{X. Xie} and \textit{H.-B. Li} [Comput. Math. Appl. 79, No. 12, 3289--3296 (2020; Zbl 1452.65054)] provided conditions for the invertibility of \(\mathcal{K}\).\N\NThe authors of the present paper investigate the invertibility of \(\mathcal{K}\) assuming that all three block diagonal matrices \(A\), \(D\), and \(E\) are possibly rank deficient. They explore the case where the leading block \(A\) has a specific nullity and present an explicit formula for \(\mathcal{K}^{-1}\).